Array shape calibration using sources in known directions

Sensor position uncertainty can severely degrade the performance of array processing. A new calibration method is presented for direction-finding with sensor position uncertainties, which is based on eigenvalue decomposition of the sample covariance matrix and three or more non-disjoint sources in known directions. The method can be applied to arbitrary arrays, including linear arrays, requires less computations and suits for low SNR cases. Computer simulations are presented to illustrate the performance of the proposed method.<<ETX>>